Part 6 · 2 chapters · ~12 min

Decision-Making Under Uncertainty

Widening options, expected value with rough numbers, reversibility, premortems, decision journals, separating decision quality from outcome quality, disagree and commit, and deciding with the right people at the right speed.

12

A decision process

Uncertainty is normal. The aim is not to be right every time; it is to decide well with what you know and learn quickly from what happens.

DECIDING UNDER UNCERTAINTY
expected value, reversibility and a record you can learn from
optionsA, B, do nothingexpected valueprobability × payoffreversibilityone-way or two-way doorpremortemimagine it failed: why?deciderecord reasons + confidencereview lateroutcome vs reasoning
swipe the figure sideways, or tap expand for full screen
1/5
options
Always include "do nothing" and at least one option you did not start with. Most bad decisions come from considering too few options.
at least three options, including doing nothingnarrow framing is the common failure
13

Decision records

code
decision record
date        2026-10-06
decision    run self-serve letters on the existing signing service; do not build our own HSM integration
options     A existing service (chosen) · B own HSM integration · C third-party e-signature vendor · D do nothing
reasons     A meets capacity with the platform team's Q4 increase; B costs ~3 months; C raises data residency questions
confidence  75% that capacity is sufficient through Q2
revisit if  month-end queue time > 10 min, or a third country launches
owner       eng lead, ledger team